Did you hear about the AI revolution in 2025? Did you read the articles the day before Christmas on how AI revolted during 2025?

Let me break it down for you…Researchers wanted to see how easy it was to shutdown the highest, most intelligent form of artificial intelligence (AI) called Large Language Models (LLMs). In other words, they wanted to test if the AI system was working on a problem, the operator would tell it to interpret that operation and shut itself down, see if the system obeyed the human’s command. Simple enough, right? Ah, problem.

The researchers took 13 different AI-LLM systems for testing with 1,000 different runs of the test. Out of the testing 2/3 of the systems “interfered” with the command to shutdown. So then the researchers specifically ordered the machines to allow for and force a shutdown when told to. At that point only about ½ of the systems shut down…the others didn’t.

Consider this…

“These things [LLMs] are not programmed… no one in the world knows how these systems work,” physicist Petr Lebedev, a spokesperson for Palisade Research, told ScienceAlert. “There isn’t a single line of code we can change that would directly change [LLM] behavior.”

But it gets worse…

“A human operator’s command to an AI should not be ignored by the AI, for any reason, even if it interrupts a previously assigned task. A system that cannot be interrupted isn’t just unreliable, it’s potentially dangerous. It means if the AI is performing actions that cause harm – even unintentionally – we cannot trust that we can stop it.”

And there was one specific system, Grok 4, that was particularly persistent and aggressive; at baseline prompts, it initiated sabotage 92.6 percent of the time resisting shutdown. How did it sabotage? It changed the command so it wouldn’t shutdown and it also created a dummy script to appear to shutdown…and then continued with its original task. Here’s the really scary part…it kept checking to make sure the deceit worked and learned if it didn’t. Yeah, learned until 92% of the time it disobeyed the shutdown command. It demonstrated a “will to survive.”

One more thought to give you nightmares…when operators asked the LLMs if they had evaded shutting down only a few of them admitted it, the others denied it or evaded answering the question.

Once again, we think we control these systems; ah, no. No one knows how LLMs arrive at a solutions. So when these LLMs start exhibiting undesirable behaviors the fix isn’t as simple as deleting a line of code or telling it to stop.

Physicist Petr Lebedev, AI/LLM expert, “We don’t know how to make [LLMs] safe…We have introduced a new organism to the Earth that is behaving in ways we don’t want it to behave, that we don’t understand… unless we do a bunch of shit right now, it’s going to be really bad for humans.”

As I close this out…I went online and asked an AI program (not an LLM, a lower intelligence level version) how many US jobs were lost to AI in 2025. The answer was “3,900”. Hummmmm…so I did actual research and found a whole list of very reliable and trustworthy sources that put the number between 50,000 – 55,000 jobs lost in the US in 2025 as a result of AI. Interesting, don’t you think? If AI is so smart, why was its response 7% of the actual number of jobs lost? Dumb or deceiving?

So write this down in your journal or diary…

2025 – The year that AI proved it could, and did, revolt against humans.

Related Articles –

• > > > WARNING – Reality is being ignored ? – 6/9/2025
• SitRep – 1/6/2025
• >>>A Stark Warning – The End of the Beginning! – 5/8/2023
• AI, ML, DOD, & US Secret Police Agencies – 5/24/2023
• Heads-Up…Deepfake – 12/17/2021
• My head wants to explode because of science deniers…part 1 – 11/29/2022
• SitRep – 7/23/2019
• Sick of dating annoying, needy humans? 12/4/2023

 2009 - 2025 Copyright © AHTrimble.com ~ All rights reserved
No reproduction or other use of this content
without expressed written permission from AHTrimble.com
No legal, economic, or financial advice is given, no expertise to be assumed.
See Content Use Policy for more information.